County lending volume and charge-off rate

Get SBA 7(a) small-business lending aggregates for a single US county. The example uses Los Angeles County, CA (FIPS 06037) — large, dense coverage, and not affected by the Connecticut FIPS schema gap that makes CT counties unusable in v1.

Problem

You want to know how active small-business lending is in a given county — how many loans, how much money, how big is the average loan, and what fraction is being charged off.

curl — latest quarter

curl -H "X-API-Key: $THESMA_API_KEY" \
  "https://api.thesma.dev/v1/us/sba/counties/06037/lending" \
  | jq '.data | {period, loan_count, total_amount, avg_loan_size, median_loan_size, charge_off_rate, jobs_supported}'

Python SDK

from thesma import ThesmaClient

client = ThesmaClient()

lending = client.sba.counties("06037").lending()

print(f"LA County — {lending.data.period}")
print(f"  loans:        {lending.data.loan_count:,}")
print(f"  volume:       ${lending.data.total_amount:,}")
print(f"  avg size:     ${lending.data.avg_loan_size:,.0f}")
print(f"  charge-off:   {lending.data.charge_off_rate:.1%}")
print(f"  jobs supported: {lending.data.jobs_supported:,}")

Expected response shape

{
  "data": {
    "county_fips": "06037",
    "county_name": "Los Angeles County, CA",
    "state_fips": "06",
    "period": "2025-Q4",
    "loan_count": 3842,
    "total_amount": 2187000000,
    "avg_loan_size": 569000,
    "median_loan_size": 245000,
    "guaranteed_amount": 1640000000,
    "avg_guarantee_pct": 0.75,
    "jobs_supported": 24180,
    "charge_off_count": 119,
    "charge_off_rate": 0.031,
    "charge_off_amount": 67800000
  },
  "metadata": {
    "source": "SBA 7(a)",
    "cadence": "quarterly",
    "latest_period": "2025-Q4"
  }
}

Values drift quarter over quarter as new SBA data lands. Field types and names are stable.

Historical time series

The default response is the latest quarter. For a time series, pass from_period and to_period:

curl -H "X-API-Key: $THESMA_API_KEY" \
  "https://api.thesma.dev/v1/us/sba/counties/06037/lending?from_period=2020-Q1&to_period=2025-Q4" \
  | jq '.data[] | {period, loan_count, charge_off_rate}'

Stable county targets

FIPS	County
`06037`	Los Angeles County, CA
`17031`	Cook County, IL (Chicago)
`48201`	Harris County, TX (Houston)
`36061`	New York County, NY (Manhattan)
`04013`	Maricopa County, AZ (Phoenix)

Avoid all Connecticut counties in v1 examples. The 2022 FIPS schema change left ~6,408 CT loans unresolved at the county level — aggregates for CT counties are missing loans that exist in state-level aggregates. See US SBA → Known limitations.

Gotchas

Quarterly cadence. Unlike BLS CES (monthly), SBA aggregates are quarterly. Periods are formatted YYYY-Qq (e.g., 2025-Q4).
Charge-off rate is dollar-weighted. charge_off_rate = charge_off_amount / total_amount, not charge_off_count / loan_count. Large single-loan charge-offs move the rate more than small ones.
Jobs-supported is self-reported by borrowers. SBA publishes it; Thesma passes it through unchanged. Treat it as directional.
County name may include a state suffix (e.g., "Los Angeles County, CA"). Parse on the comma if you need the bare county name.